Input Data Options
special purpose data requiring pre-processing (0)
tomographic data (1)
segmented data (2)
burn data (3)
medial axis data (4)
throat data (5)
fluid data (6)
Enter choice: 2
Data Processing Options
plot image (1)
resize data (2)
disconnected volume distribution (3)
clean up segmented files (4)
burn and compute LKC medial axis/surface (5)
moment of inertia of disconnected components (6)
convert data to/from ascii format (7)
2-point covariance function (8)
tomographic/segmented data covariance comparison (9)
compress data in z direction (10)
count voxels in spherical/cylindrical shells (11)
set fiducial polygon exterior to grain phase (12)
overlay segmented and tomographic images (13)
pore erosion disconnectivity analysis (14)
Enter choice: 4
Files assumed to
be labelled basename.ext,
basename
is limited to 255 characters maximum,
ext
is a numerical designator lying between 000 and 999
It is assumed that
the files have consecutively numbered extensions
ie. 000 -> 056, or 021 -> 049
Enter basename for input segmented
files: ../seg/sw
Are files compressed? [y,n]: y
Enter first and last slice of data to use: 1 256
Input data can be inverted to
compute
burn/medial axis of
grain space rather than void space.
Invert data (y,n(dflt))?: n
Whether or
not you want to invert depends on your data. In this case,
the water phase has been identified as the void(phase 0) and berea
rock(phase 1). It is customary for many
algorithms
to assume the grain
space to be marked as phase 1 (and void
space as
0) so in this particular
example we have no need to invert. There might be
some cases where the
actual
pore space is identified as phase 1, because, for instance,
it is filled with fluid that has higher X-ray attenuating factor than
the actual grain.
Fiducial polygon generation methods
NONE (N)
MANUAL (M)
AUTOMATIC (A)
Enter method: N
Correct for ring artifacts? (y,n(dflt)): n
Ring
artifacts
show up from time to time in the microtomography images
and are the
imaging techique artifact - radial rings can be evident
in your
segmented
image. The correction is trying to reconstruct the
image from
the neighboring voxels.
The material/void boundary can be
lightly
smoothed.
Available options are
0) no conversion
or convert those boundary voxels having
1) exactly one neighbor of the same type
2) less than a majority of neighbors of the same type
Enter choice (0(dflt),1,2): 0
One
doesn't
know what is beyond the boundary. We prefer no conversion
of the
segmentation
decisions on the boundary. However, depending on the
problem you
might choose to change the boundary voxels segmented values
if all but
one neighbor are of the opposite phase(1) or if the majority
of the
neghboring
voxels is of the opposite phase.
Isolated clusters of grain and/or
pore
voxels up to a specifed size
can be assumed to be misidentified
and converted to the opposite material type.
Convert isolated grain clusters?
(y,n):
y
Enter maximum allowed size (number
of voxels)
for convertible isolated grain cluster: 1000
Unless
they
are touching the boundary, isolated grains that "float" in
void space
are unphysical.
Convert isolated pore
clusters?
(y,n): y
Enter maximum allowed size (number
of voxels)
for convertible isolated pore cluster: 1000
Maximum
allowed
size of the isolated, disconnected pieces of either phase
are up to
the human observer and the data analysis needs. The isolated clusters
may produce
noise in, e.g. medial axis computation, that one would like to
avoid. In
some bigger sample you might want the increase the above maximum
allowed sizes
since even 10000 voxels might not mean much...
Enter basename for output segmented
files: ../c_seg/sw
Are files to be compressed? [y,n]:
y
Prepare raster files of corrected
segmented
image (y,n(dflt)): y
Enter basename for raster files:
../c_seg/sw
Merge raster files of segmented image (y,n(dflt)): n